The present results demonstrate that certain morphometric brain abnormalities associated with AD are less salient in very-old patients than in young-old patients despite similar levels of global cognitive impairment in the 2 groups. When compared with their respective age-appropriate HC groups, very-old patients showed less severe cortical thinning than young-old patients in the left PCC, right lateral temporal cortex, and bilateral parietal cortex and in overall cortical thickness averaged across all ROIs. This effect is partially explained by an age-related decrease in cortical thickness in these brain regions in the HC participants (). Although the AD groups had similar overall cortical thickness, the very-old patients' age-appropriate standard scores were less abnormal than those of the young-old patients because of reduced and more variable cortical thickness in the very-old HC participants. Thus, cortical thickness reduction that can be attributed to AD in these brain regions is more difficult to detect in very-old patients with AD than in young-old patients with AD.
The observed pattern of age-related differences in abnormalities in cortical thickness is reflected in the pattern of age-related differences in the severity of cognitive abnormalities exhibited by the AD groups. Consistent with previous findings,5,24
the present results showed that very-old patients with AD were less impaired (i.e., were less abnormal) than young-old patients with AD in the domains of immediate memory, attention/processing speed, and executive function. As with cortical thickness measures, the reduced degree of abnormality in these cognitive domains can be at least partially explained by an age-related decrease in performance in HC participants (). Very-old HC participants scored significantly lower than young-old HC participants on a number of measures in these domains (e.g., Digit Symbol Substitution, Trail Making Test Part B, Digit Span–Forward). Similar age-related decrements were reported previously and were attributed, in part, to deterioration of the frontoparietal cortical systems involved in attention and response selection.25,26
Not all brain regions usually affected by AD showed differential levels of volumetric or cortical thickness abnormality in very-old and young-old patients with AD. For example, the AD groups did not significantly differ in degree of abnormality in hippocampal volume or medial temporal cortical thickness, corroborating previous results.27
Although atrophy of the hippocampus and medial temporal gyrus increased with age in the HC participants, the degree of difference between the HC and AD groups remained large. Consistent with this finding, the AD groups did not differ in the degree of abnormality on delayed recall or savings measures (although this result approached significance) that are largely mediated by the hippocampus and related medial temporal lobe (MTL) structures. These results suggest that early and severe atrophy of the hippocampus and MTL cortex due to AD greatly eclipses normal age-related changes and allows atrophy in these regions to be a salient marker of AD regardless of the elderly patient's age.
It has been proposed that imaging or other types of biomarkers are critically needed to detect AD early so that interventions can be applied before the disease significantly diminishes cognitive function.28
Previous studies have shown that MRI-derived indices of atrophy have high sensitivity and specificity for detecting disease in symptomatic patients and, therefore, have the potential to be effective biomarkers. However, our results suggest that it is critical to consider the effect of age in application of MRI-derived markers because they may be better able to distinguish patients with AD from healthy elderly individuals in the young-old than in the very-old groups. A similar loss of saliency may occur with aging for other biomarkers (e.g., tau and Aβ in CSF), given the overlap that occurs in AD and the aging process. Indeed, some evidence suggests that AD-related increases in the densities of neuritic plaques and neurofibrillary tangles that underlie some biomarkers are less profound in very-old than in young-old individuals.29,30
Some limitations of the present study should be noted. First, the cross-sectional nature of the study limits the ability to determine whether rate of change in hippocampal volume or cortical thickness is similar in very-old and young-old patients with AD and whether abnormality in rate of change is influenced by age. The results need to be replicated longitudinally to address this issue and to directly compare rate of change in regional brain volumes against rates of cognitive decline. Second, histopathologic verification of disease is not available so it is possible that some participants have a disorder other than AD or have AD with comorbid pathology that contributes to cognitive and neuroimaging presentations. The prevalence of vascular pathology in AD, for example, has been reported to be roughly 30%–60%.28,31
Although ADNI exclusionary criteria ensure a low prevalence of vascular risk factors, the impact of white matter changes on the pattern of cognitive and regional brain changes in AD across different age groups may help explain some of the observed differences in cognitive profiles. Third, impaired initial learning can lead to an inflated savings score (percentage retained) in some cases; thus, results from this measure must be interpreted with caution. For example, an individual who learns only one item on list learning and then recalls only that same item after a delay would have 100% savings, despite poor overall memory performance. To reduce the chance of significant inflating in our savings variable, we excluded any outliers (≥3 SD from the overall group mean). Fourth, other brain variables may contribute to the different cognitive deficit profiles seen in the 2 groups. For example, normal aging is associated with mild brain atrophy on structural MRI,6
decreased hemodynamic response on fMRI,32
reduced synaptic density,33
increased white matter hyperintensities,6,34,35
and a subclinical accumulation of neuritic plaques and neurofibrillary tangles in MTL regions.36
These brain changes are accompanied by age-related declines in information-processing speed, executive functions, and efficiency of learning and recall3,4,35,37
and may help explain the observed differences in severity of cognitive abnormalities in very-old and young-old patients with AD. These possibilities will be addressed in future studies.
Overall, the present results indicate that overlap between normal and AD-related MRI-based morphometric changes is greater in the very old than in the young old. Thus, the typical pattern of AD-related morphometric changes seen in the young old is less salient in the very old. A similar loss of saliency in cognitive profiles is evident because of greater overlap in normal and AD-related decline in executive functions, attention/psychomotor processing, and immediate memory in very-old than in young-old individuals. These results highlight the possibility that mild cases of AD in the very old may go undetected and underscore the importance of interpreting neuropsychological test performance and morphometric changes (i.e., atrophy) in reference to the individual's age. A clarification of how the presentation of AD changes with age may enhance our ability to detect early AD in the very old, one of the fastest growing segments of the population.38
Indeed, enhanced detection is crucial for early application of interventions that may slow the disease process, thus preserving cognitive status, functional independence, and quality of life.